The FDA recognizes osteoarthritis as a serious disease that has few effective treatments and none accepted to reduce its structural progression. Barriers to developing these therapies include failure to enrich study samples with people likely to progress, an absence of standard definitions of disease progression, and a lack of structural measures that reliably predict clinically meaningful outcomes. Magnetic resonance (MR) imaging has great potential to address the gaps impeding drug testing. However, no-one has developed an MR-based quantitative composite outcome that reflects multiple structural aspects (?whole-knee?) of knee osteoarthritis (KOA) progression, is sensitive to change, and is prognostic of clinically meaningful outcomes. To tackle these barriers, we developed and tested parsimonious semi-automated MR analysis methods to quantify knee cartilage damage, bone marrow lesions (BMLs), and effusion-synovitis. Next, we used an iterative process to combine these measures and validated 2 composite metrics reflecting separate domains. The first is cumulative damage, which reflects whole knee cartilage damage. The second is disease activity, which is based on BML and effusion-synovitis volumes and relates to joint symptoms. These composite metrics were accepted into the FDA CDER Biomarker Qualification Program with a proposed prognostic context of use for the enrichment of KOA clinical trials. The critical next steps are to determine the thresholds for KOA cumulative damage and disease activity metrics (cross-sectional and longitudinal change) that optimize their prognostic performance in predicting knee arthroplasty. We will also explore how these composite metrics perform in predicting knee arthroplasty compared with standard structural measures (e.g., joint space width, cartilage thickness) and each component measure of the composite metrics (13 measures of cartilage damage, BML volume, and effusion-synovitis). We will also explore the consistency of these prognostic measures across varying acquisition periods (1, 2, and 3 years). We will accomplish this by leveraging the resources of the Osteoarthritis Initiative, which was designed to develop KOA biomarkers. Specifically, we will use a matched nested case-control design with a sample of case knees that received a total knee arthroplasty and then perform ROC analyses. This proposal will further the development of cumulative damage and disease activity as prognostic enrichment imaging biomarkers for long-term KOA progression. These prognostic biomarkers will be made publicly available to enrich study samples for a clinical trial. They may also be used in early phase trials to demonstrate the therapeutic potential of an intervention to advance to Phase 3 trials.